Abstract
The in vivo P50 (P50iv) provides a useful index of haemoglobin-oxygen affinity and is calculated according to software algorithms incorporated into commercial blood gas analysers. These algorithms are known to be inaccurate at high haemoglobin saturation (SpO2 >97%) although just how inaccurate has not been documented. This study examines the arterial blood gas profiles of patients admitted to a busy secondary referral Intensive Care Unit and stratifies them according to haemoglobin saturation in order to quantify the accuracy and potential clinical utility of the Siggaard-Andersen algorithm (SAA) for assessing P50iv in blood with SpO2 >90%. Sicker patients, as identified by plasma pH <7.35 or [lactate] >2.0 mmol/l, were substratified and the SAA assessed as before. In both groups, the results show not only that the SAA is completely unreliable above 97% saturation, a fact acknowledged by Siggaard-Andersen in 1984, but it is also inaccurate in the range 92% ≤ SpO2 ≤ 97%, thus rendering P50iv calculations suspect in 90% of the patients in each of the study groups.